Critical Flaws: Hugging Face Diffusers' Custom Code Safeguards Bypassed by Malicious Models
Three CVEs in Hugging Face Diffusers allow malicious model repos to execute arbitrary code, bypassing custom code safeguards.
20 Hours to Catastrophe: How Hackers Exploited a Critical Langflow CVE in Under a Day
Threat actors rapidly exploited a critical Langflow CVE within 20 hours, highlighting urgent AI supply chain security risks.